JP5008748B2 - Search method, integrated search server, and computer program - Google Patents

Description

  The present invention relates to a search method, an integrated search server, and a computer program.

  In the full text search service, a search server analyzes file data stored in a computer system and creates a search index in advance. The search server provides a search service to the user using the search index. The user transmits a search query for searching for a file to be acquired to the search server, and accesses the target file based on the search result. Since the number of files stored in a computer system is increasing year by year, the full-text search service has become an important service for users.

  By the way, when there are a plurality of search servers, the user needs to issue a search query to each search server individually and acquire the search results from each search server individually. For this reason, user convenience is poor.

  In recent years, therefore, an integrated search service has been provided in which search results from each search server can be acquired in an integrated manner by issuing a search request only once to a plurality of independent search servers.

  For example, OpenSearch, a specification for integrated search, has been released, and an integrated search service using the specification is provided. In the integrated search service, each search server is operated independently. On the other hand, each search server can accept a search request based on a unified standard interface such as OpenSearch. This enables an integrated search in which a plurality of search servers are loosely coupled. In an integrated search that is loosely coupled, the search algorithm used by each search server or the search trigger for the search index is different.

  On the other hand, there is also a form that provides a tightly coupled integrated search service by operating a plurality of search servers integrally. In a tightly coupled integrated search service, each search server uses the same search algorithm, and the search index is also updated uniformly in the system. An integrated search service in which a plurality of search servers are closely coupled can be viewed as a single search server.

  Further, a search server having a function of excluding duplicate contents from search results is also known. Specifically, the search server detects a duplicate entry based on a hash value generated from each entry of the search result, and deletes the duplicate entry from the search result (Patent Document 1).

US Pat. No. 7,366,718

  The technique described in the above-mentioned document can only delete duplicate entries in each search server, and it is practically difficult to detect duplicate entries for integrated search results obtained by integrating search results from a plurality of search servers. There is.

  This is because in the case of an integrated search service in which a plurality of search servers are loosely coupled, hash algorithms used to detect duplicate entries may be different for each search server. If the hash algorithm used by each search server is different, it is extremely difficult to detect duplicate entries based on the hash value. Therefore, with the technique described in the document, it is not possible to detect duplicate entries included in the integrated search result in a system in which a plurality of search servers are loosely coupled.

  The above-described problem is caused by the difficulty in unifying the hash algorithm used for detecting duplicate entries among the search servers. Various hash algorithms already exist, and various new hash algorithms will appear and be implemented in search servers. Furthermore, the requirements for the hash algorithm are different for each search server.

  For the above reasons, it is practically difficult to unify the hash algorithm among a plurality of search servers that are operated independently. Therefore, the prior art described in the above document cannot be applied to an integrated search service in which a plurality of search servers are loosely coupled.

  Therefore, an object of the present invention is to provide a search method, an integrated search server, and a computer program that can detect and eliminate duplicate data from each search result in a system in which a plurality of search servers are loosely coupled. . Further objects of the present invention will become clear from the description of the embodiments described later.

  In order to solve the above-described problem, a search method according to the first aspect of the present invention is a search method using a computer system including a plurality of search servers, and the computer system operates independently from each other. When the integrated search server included in the plurality of search servers receives the integrated search request for searching the plurality of predetermined search servers included in the plurality of search servers, The search servers that can be used in common by each search server determine duplicate detection information for detecting duplicates, issue a search request corresponding to the integrated search request to each predetermined search server, and each predetermined search server Searches the data group in charge based on the search request, and creates a duplicate detection value for detecting duplicates created using the determined duplicate detection information in the search result. The integrated search server detects duplicate data based on each duplicate detection value from the search results received from each predetermined search server, and is detected from each search result. The integrated search result is created by removing the duplicate data, and the integrated search result is provided to the issuer of the integrated search request.

  In the second aspect, in the first aspect, each predetermined search server stores in advance a duplicate detection value for each of a plurality of pieces of duplicate detection information for the data group in charge, and each of the stored duplicate detection values is stored in advance. Among them, the duplicate detection value corresponding to the duplicate detection information determined by the federated search server is included in the search result and transmitted to the federated search server.

  In the third aspect, in the second aspect, each predetermined search server creates a duplicate detection value for each of a plurality of duplicate detection information when updating a search index used to search a data group in charge. And save.

  In the fourth aspect, in the first aspect, the integrated search server acquires information about duplicate detection information that can be used in each predetermined search server from each predetermined search server and holds the information, and issues an integrated search request. When received, the duplication detection information that can be used in common by each predetermined search server is determined based on the information relating to each held duplication detection information.

  In a fifth aspect, in the first aspect, the integrated search server acquires information on duplicate detection information that can be used in each predetermined search server from each predetermined search server when the computer system is constructed. When the integrated search request is received, duplicate detection information that can be commonly used by each predetermined search server is determined based on the held information regarding each duplicate detection information.

  In a sixth aspect, in the first aspect, when each predetermined search server receives a search request from the integrated search server, it creates a duplicate detection value based on the determined duplicate detection information and searches for the duplicate detection value. Included in the result and sent to the integrated search server.

  In a seventh aspect, in the first aspect, the duplicate detection information is a hash algorithm, and the duplicate detection value is a hash value.

  The present invention is an integrated search server for searching using a computer system configured by loosely coupling a plurality of search servers that operate independently, or a computer program for causing a computer to function as an integrated search server. I can understand. Note that combinations other than the above-described viewpoints are also included in the scope of the present invention. The computer program can be distributed via a communication medium or a recording medium.

FIG. 1 is a diagram showing an overall configuration of a computer system. FIG. 2 is a diagram illustrating a hardware configuration of the search server. FIG. 3 is a diagram showing a configuration of a computer program stored in the search server. FIG. 4 is a diagram illustrating a configuration of a table stored in the search server. FIG. 5 is a diagram illustrating a hardware configuration of the file server. FIG. 6 is a diagram illustrating a hardware configuration of the client machine. FIG. 7 is a diagram schematically showing a series of integrated search processes. FIG. 8 shows a table for managing files registered in the search index. FIG. 9 shows a table for managing the search index. FIG. 10 shows a table for managing the search server. FIG. 11 shows a table for temporarily storing the integrated search results. FIG. 12 shows a configuration example of the integrated search request parameter. FIG. 13 shows a configuration example of response parameters of the integrated search result. FIG. 14 shows a configuration example of the hash algorithm inquiry request parameter. FIG. 15 shows an example of the configuration of response parameters for a hash algorithm query. FIG. 16 shows a configuration example of the search request parameter. FIG. 17 shows a configuration example of the response parameter of the search result. FIG. 18 is a flowchart showing integrated search request processing. FIG. 19 is a flowchart showing an integrated search process. FIG. 20 is a flowchart following FIG. FIG. 21 is a flowchart showing processing for responding to a hash algorithm query. FIG. 22 is a flowchart showing a process of searching and responding with a search result. FIG. 23 is a flowchart showing processing for updating a search index. FIG. 24 is a flowchart following FIG. FIG. 25 relates to the second embodiment and shows a configuration example of a table for managing the search server. FIG. 26 is a flowchart showing a part of the integrated search process. FIG. 27 shows a configuration example of a computer program included in the search server according to the third embodiment. FIG. 28 is a flowchart showing processing for negotiating a hash algorithm in advance. FIG. 29 is a flowchart showing search response processing according to the fourth embodiment. FIG. 30 is a flowchart showing a part of the federated search process according to the fifth embodiment. FIG. 31 relates to the sixth embodiment and shows a table for temporarily storing the integrated search results. FIG. 32 is a flowchart showing a part of the integrated search process. FIG. 33 is a flowchart illustrating processing for notifying a search server of duplicate entries according to the seventh embodiment.

  1100: Integrated search server / search server, 1200, 1300: Search server, 2100, 2200, 2300: File server, 3100, 3200, 3300: Client machine, 1124: Search control program, 1125: Integrated search control program, 5100: Integration Search control unit, 5110, 5210, 5310: Search control unit, 5120, 5220, 5320: Search index

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. In the present embodiment, a processing method for the search server to detect and eliminate duplicate contents from the integrated search result will be described. As described in detail below, in the present embodiment, a hash algorithm used in each search server that performs a search is determined in advance, and a hash value calculated by the determined hash algorithm is included in the search result, and the integrated search server Send to. The federated search server detects and removes duplicate entries using the hash value.

  FIG. 1 is an explanatory diagram illustrating a system configuration according to this embodiment. A plurality of search servers 1100, 1200, 1300, a plurality of file servers 2100, 2200, 2300, and a plurality of client machines 3100, 3200, 3300 are connected via the communication network 100. Further, a server 7000 for distributing computer programs may be connected to the communication network 100.

  In this system, a corresponding search server creates a search index for data stored in each file server. Each search server provides a search service for a file on the file server to the client machine using the search index. Further, the search server also provides an integrated search service that provides the client machine with search results from a plurality of search servers.

  The specific service contents are as follows. First, the client machine can register a file (data file) in the file server. The file server stores and stores the registered file in an external storage device connected to the file server. The search server acquires a file stored in the file server by crawling, and creates a search index. The search server stores and stores the search index in an external storage device connected to the search server.

  The client machine can transmit a search request by specifying a search query to the search server. The search server extracts a file that matches the search query condition using the search index of the search server, and provides the search result to the client machine.

  Furthermore, the client machine can send an integrated search request to the search server by specifying a search query. The search server extracts files that match the search query conditions using a search index of the search server. Further, the search server also transmits a search request to other search servers that can perform integrated search, and provides the search result returned from each search server to the client machine as an integrated search result.

  The client machine can select an access target file based on the integrated search result. The client machine can access the file stored in the file server by using the file access file path name stored in the integrated search result.

  In FIG. 1, three types of devices, a search server, a file server, and a client machine, are shown as separate devices. For example, any two or all three of the three types of devices may be configured as one computer device.

  The program distribution server 7000 is a device that distributes a program such as a hash algorithm to a search server, for example. For example, the program distribution server may be integrated with a file server or a search server and realized in one computer apparatus.

  Furthermore, the connection form by the communication network 100 may be an Internet connection or an intranet connection by a local area network.

  FIG. 2 is an explanatory diagram illustrating the hardware configuration of the search server 1100. In this embodiment, of the three search servers 1100, 1200, and 1300, the search server 1100 is a reception window for the integrated search service. That is, the search server 1100 is an “integrated search server” for providing an integrated search service to a client machine, and is also a “predetermined search server” for searching according to a search request.

  The search server 1100 includes, for example, a processor 1110, a memory 1120, an external storage device interface (hereinafter referred to as I / F) 1130, a network I / F 1140, and a bus 1150 that connects these 1110, 1120, 1130, and 1140. Consists of including.

  The processor 1110 executes a computer program (hereinafter referred to as a program). The memory 1120 stores programs 1121-1125 and tables 4100-4400, which will be described later. The external storage I / F 1130 is a communication circuit for accessing the external storage device 1160. The network I / F 1140 is a communication circuit for accessing other devices (such as a file server and a client machine) via the communication network 100.

  FIG. 3 shows program contents stored in the memory 1120. In the memory 1120, for example, an external storage device I / F program 1121, a network I / F program 1122, a data management program 1123, a search control program 1124, and an integrated search control program 1125 are stored.

  The external storage device I / F program 1121 controls the external storage device I / F 1130. The network I / F program 1122 controls the network I / F 1140. The data management program 1123 provides a file system or database used for managing stored data in the search server 1100. The search control program 1124 provides a search service in the search server 1100. The integrated search control program 1125 provides an integrated search service in the search server 1100.

  FIG. 4 shows the contents of a table (management data) stored in the memory 1120. In the memory 1120, for example, a search index registration file management table 4100, a search index management table 4200, a search server management table 4300, and an integrated search result temporary storage table 4400 are stored.

  The search index registration file management table 4100 is a table used by the search control program 1124, and manages files registered in the search index. The search index management table 4200 is a table for managing search indexes. The search server management table 4300 is a table used by the integrated search control program 1125, and manages each search server included in the integrated search system. The integrated search result temporary storage table 4400 is a table used by the integrated search control program 1125, and temporarily stores the results of the integrated search.

  Returning to FIG. The search control program 1124 includes therein a search index management subprogram 1171, a search reception subprogram 1172, a hash algorithm response subprogram 1173, and a deduplication subprogram 1174.

  The search index management subprogram 1171 performs processing necessary to manage search index data. Specifically, the search index management subprogram 1171 performs crawling processing on the file server 3100 that stores the file data that is the search target of the search server 1100, and generates search index data as necessary. , Update, or delete. The search index management subprogram 1171 manages the substance of the search index data using the data management program 1123.

  The search reception subprogram 1172 receives a search request specifying a search query from the client machine. The search reception subprogram 1172 searches for a file that matches the search condition, and performs a process of responding the search result to the client machine. In this embodiment, the search reception subprogram 1172 performs search processing using search index data separately created by the search index management subprogram 1171.

  The hash algorithm response subprogram 1173 is a program that receives a request and performs a response after receiving a request when another search server requests a hash algorithm negotiation. The hash algorithm response subprogram 1173 responds to the inquiry source with a list of hash algorithms that can be used by the search server in which the hash algorithm response program 1173 is installed. Although details will be described later, duplication in the integrated search result can be detected by instructing each search server to use a hash algorithm that can be commonly used in each search server.

  In the following description, a search server on which a subject program or table is mounted may be referred to as a self-search server.

  The deduplication subprogram 1174 detects duplicate data in the search index data managed by the search index management subprogram 1171 of its own search server, and performs a process of deleting the duplicate data as necessary. That is, the deduplication subprogram 1174 eliminates duplicate data stored in one search server.

  In order to detect duplicate data, a hash algorithm described later is used. The deduplication subprogram 1174 determines whether any given data in the search index data is identical to other data based on the hash value calculated by the hash algorithm.

  The integrated search control program 1125 includes therein an integrated search reception subprogram 1175, a hash algorithm negotiation subprogram 1176, and an integrated search result deduplication subprogram 1177.

  When the integrated search reception subprogram 1175 receives an integrated search request specifying a search query from a client machine, the integrated search reception subprogram 1175 also uses a plurality of other search servers capable of integrated search to search for a file that matches the search condition. The integrated search reception subprogram 1175 collects the search results from each search server and transmits them to the client machine as the integrated search results. The integrated search reception subprogram 1175 uses the search server management table 4300 to select search servers that can perform integrated search.

  The hash algorithm negotiation subprogram 1176 is a search server group that can perform an integrated search on a hash algorithm used to eliminate duplicate contents from the integrated search result when the integrated search reception subprogram 1175 receives an integrated search request. Necessary processing to negotiate and agree with Specific processing contents will be described later.

  The integrated search result deduplication subprogram 1177 detects duplicate data from search result data acquired from a search server group capable of integrated search, and performs a process of deleting the duplicate data as necessary. The integrated search result deduplication subprogram 1177 uses a hash algorithm negotiated with another search server group in order to detect duplicate data. The integrated search result deduplication subprogram 1177 determines whether any data in the search result data is the same as other data by using the hash algorithm.

  The search index registration file management table 4100, the search index management table 4200, the search server management table 4300, and the search result temporary storage table for integration 4400 will be described later.

  Other search servers 1200 and 1300 have the same configuration as search server 1100 except that they do not have a configuration related to integrated search (integrated search control program 1125, search server management table 4300, and integrated search result temporary storage table 4400). Therefore, the explanation is omitted.

  FIG. 5 is an explanatory diagram illustrating the hardware configuration of the file server 2100. The file server 2100 includes, for example, a processor 2110 that executes a program, a memory 2120 that temporarily stores the program and data, an external storage device I / F 2130 that accesses the external storage device 2160, and the network 100. A network I / F 2140 for communicating with other devices (such as a search server) and a bus 2150 for connecting them are provided.

  In the memory 2120, for example, an external storage device I / F program 2121, a network I / F program 2122, a file sharing service program 2123, and a file management program 2124 are stored.

  The external storage device I / F program 2121 controls the external storage device I / F 2130. The network I / F program 2122 controls the network I / F 2140. The file sharing service program 2123 manages the file sharing service provided from the file server 2100. The file management program 2124 manages the files stored in the file server 2100.

  The file sharing service program 2123 uses the file management program 2124 to manage shared files. The search server or client machine can access the shared file stored in the file server 2100 by using the file sharing service program 2123.

  Since the other file servers 2200 and 2300 have the same configuration as the file server 2100 described here, the description thereof is omitted.

  FIG. 6 is an explanatory diagram illustrating the hardware configuration of the client machine 3100. The client machine 3100 is connected to, for example, a processor 3110 that executes a program, a memory 3120 that temporarily stores the program and data, and an external storage device I / F 3130 that accesses the external storage device 3160 via a network. A network I / F 3140 for accessing other devices and a bus 3150 for connecting them are provided.

  For example, an external storage device I / F program 3121, a network I / F program 3122, a file management program 3123, a search service client program 3124, and a file sharing service client program 3125 are stored in the memory 3120.

  The external storage device I / F program 3121 controls the external storage device I / F 3130. The network I / F program 3122 controls the network I / F 3140. The file management program 3123 provides a file system for managing files stored in the client machine 3100. The search service client program 3124 is a program for using the search service provided by the search server 1100 and the integrated search service. The file sharing service client program 3125 is a program for using the file sharing service provided by the file server 2100.

  The search service client program 3124 uses an HTTP client program (for example, a web browser) when the search service and the integrated search service use the HTTP protocol.

  The file sharing service client program 3125 uses an NFS client program when the file sharing service uses the NFS protocol. When the file sharing service uses the CIFS protocol, the file sharing service client program 3125 uses the CIFS client program. Alternatively, when the file sharing service uses the HTTP protocol, the file sharing service client program 3125 uses an HTTP client program (such as a Web browser).

  Since the other client machines 3200 and 3300 have the same configuration as the client machine 3100, description thereof is omitted.

  FIG. 7 schematically shows the operation of the entire system when an integrated search request is issued from the client machine 3100 to the search server 1100. In FIG. 7, a series of processes such as issuing an integrated search request, searching at each search server, obtaining a search result from each search server, and providing an integrated search result are described in nine steps. Hereinafter, the step may be abbreviated as “S”.

  In the following description, the same reference numeral 1100 is assigned to the search server 1100 as the “integrated search server” that executes the integrated search processing and the search server 1100 as the “predetermined search server” to search according to the integrated search request. For example, in a sentence such as “The search server 1100 that has received an integrated search request requests a search from each of the search servers 1100, 1200, and 1300.” An integrated search server that receives requests and executes integrated search processing, and mainly corresponds to the integrated search control program 1125. The “search server 1100” in the “search servers 1100, 1200, 1300” is a search server that performs a specified search and returns a result, and mainly corresponds to the search control program 1124.

  First, as S1, the client machine 3100 sends an integrated search request to the search server 1100 that provides the integrated search service. In the integrated search request, a search keyword and a search condition are specified.

  Search keywords and search conditions used for the integrated search can be specified in the same manner as search keywords and search conditions that can be accepted by a conventional general search engine. For example, a plurality of character strings may be specified as search keywords. As a search condition, a data creation date or a data last update date may be specified in an arbitrary range, or a data creator may be specified.

  In S2, the integrated search control unit 5100 in the search server 1100 that has received the integrated search request sends a hash algorithm (identification information such as a usable hash function) to the search servers 1100, 1200, and 1300 that can use the integrated search. Equivalent)). The integrated search control unit 5100 is mainly realized by the integrated search control program 1125.

  Upon receiving the integrated search request, the search server 1100 specifies a hash algorithm that can be used by the self-search server 1100, and determines whether the other search servers 1100, 1200, and 1300 can use the hash algorithm. The search servers 1200 and 1300 are inquired.

  As S3, the search control units 5110, 5210, and 5310 in the search servers 1100, 1200, and 1300 that have been queried use information other than the specified hash algorithm and whether or not the specified hash algorithm is supported. Information on possible hash algorithms is returned to the integrated search control unit 5100 that is the inquiry source. Search control units 5110, 5120, 5130 are realized by hash algorithm response subprogram 1173.

  The integrated search control unit 5100 determines a hash algorithm that can be used for the integrated search based on the response results from the search control units 5110, 5210, and 5310. In the following description, a hash algorithm that can be used for integrated search may be referred to as a common hash algorithm. If the common hash algorithm cannot be determined by a single query, the query and response may be repeatedly executed a predetermined number of times.

  In S4, the integrated search control unit 5100 sends the same search request to the search servers 1100, 1200, and 1300 that can use the integrated search. The search request includes information related to the common hash algorithm determined in the above-described process, together with the search keyword and search condition included in the integrated search request.

  As S5, the search control units 5110, 5210, and 5310 execute search processes using the search indexes 5120, 5220, and 5320 managed in the search servers 1100, 1200, and 1300, respectively. In the search process, the search keyword and the search condition specified from the integrated search control unit 5100 are used.

  In S6, the search control units 5110, 5210, and 5310 perform deduplication processing on each search result. Specifically, each of search control units 5110, 5210, and 5310 checks whether or not a plurality of entries indicating the same file are registered in the entries included in the search results.

  If a plurality of entries indicating the same file are registered, the search control units 5110, 5210, 5310 leave only one arbitrary entry and hide other entries in accordance with a predetermined duplication deletion condition. Or delete.

  A hash algorithm is used to determine whether the files are the same. Specifically, a hash function or the like is used. The search control units 5110, 5210, and 5310 generate hash values by using a hash function for each file data or for a plurality of file data to be determined whether they are the same. If the hash values match, it can be determined that the files are identical.

  In step S7, the search control units 5110, 5210, and 5310 respond to the search results obtained by removing the duplicate entries in the search servers 1100, 1200, and 1300 to the integrated search control unit 5100 of the search server 1100 that is the search request source. To do.

  Further, the search control units 5110, 5210, and 5310 also provide the integrated search control unit 5100 with information generated by using the common hash algorithm specified in S4 in addition to the search results. Specifically, the integrated search control unit 5100 is notified of a hash value generated using a hash function corresponding to the common hash algorithm.

  In step S8, the integrated search control unit 5100 generates an integrated search result based on the search result acquired from each search server, and further performs processing for eliminating duplicate entries from the integrated search result. Hereinafter, the process of removing duplicate entries from a plurality of entries included in the integrated search result may be referred to as an integrated search result deduplication process.

  The specific content of the integrated search result deduplication processing is almost the same as the content of the deduplication processing in each of the search control units 5110, 5210, and 5310 described above. Specifically, it is checked whether or not there are a plurality of entries indicating the same file data in each entry included in the integrated search result. If a plurality of entries indicating the same file are registered in the integrated search result, only one arbitrary entry is left and other entries are hidden or deleted according to a predetermined duplication deletion condition.

  A hash algorithm is used to determine whether or not a plurality of file data are the same. Specifically, the hash value calculated by the common hash algorithm provided from each search server is used. When hash values (hash values created in the search server) of a plurality of file data match, it can be determined that the file data are the same.

  Finally, in S9, the integrated search control unit 5100 responds to the client machine 3100 with the integrated search result from which duplicate entries are removed. Through the above processing, the client machine 3100 can acquire the integrated search result.

  FIG. 8 shows a configuration example of the search index registration file management table 4100. The search index registration file management table 4100 manages information related to files acquired by the search server from the file server for which the search index is to be created. Specifically, the search index registration file management table 4100 includes a file ID 4110, an acquisition source file path name 4120, target file metadata 4130, a cache storage destination 4140, and a hash algorithm (and hash value) of the target file. 4150 is managed in association with each other.

  The file ID 4110 is an identifier for uniquely identifying a file acquired from the file server. The file ID 4110 may be a serial number given by the search server 1100 or a serial number given by the file server 2100.

  The acquisition source file path name 4120 is a file path name indicating the storage destination of the target file in the file server. The search server issues a file acquisition request specifying the acquisition source file path name 4120 to the file server. Thereby, the search server can acquire a desired file from the file server.

  The target file metadata 4130 is a set of metadata associated with the target file. The metadata 4130 corresponds to, for example, a file owner, file creation date / time, file size, file access right information, etc. managed by the file server. Furthermore, information such as the latest file acquisition date and time managed by the search server can also be included in the metadata 4130.

  The cache storage location (storage location) 4140 is information indicating a storage location when the cache data of the target file is stored in the search server. Specifically, when the cache data is managed in the file format in the search server, the storage path name of the file is registered in the cache storage destination column 4140.

  The target file hash algorithm and hash value column 4150 stores information used for detecting duplication of the target file data. The column 4150 includes columns 4151 and 4153 for registering hash algorithms, and columns 4152 and 4154 for registering hash values.

  In the hash algorithm fields 4151 and 4153, identification information of hash functions used for duplicate detection is registered. In the hash algorithm fields 4151 and 4153, information for identifying a hash function such as MD5 or SHA-1 is registered. In the hash value columns 4152 and 4154, hash values created using the hash functions registered in the hash algorithm columns 4151 and 4153 are registered.

  The hash algorithm / hash value field 4150 is configured so that a plurality of pairs of hash algorithms and hash values can be registered. FIG. 8 shows an example in which two sets are registered for each file. Three or more sets may be registered. Further, the same number of sets may be registered for all the files, or the number of sets that can be registered for the set of hash algorithm and hash value may be different for each file.

  FIG. 9 shows a configuration example of the search index management table 4200. The search index management table 4200 manages search index information created by the search server. Specifically, the search index management table 4200 manages the keyword 4210 and the position information 4220 in association with each other.

  The keyword 4210 stores a character string obtained by indexing the target file. In the position information 4220, file information including the character string of the keyword 4210 is registered. The position information 4220 includes file IDs 4221 and 4224, corresponding position offsets 4222 and 4225, and weighting coefficients 4223 and 4226.

  File IDs 4221 and 4224 register information for identifying a file in which a character string of a keyword appears. The file ID registered in the file ID 4110 column of the search index registration file management table 4100 is registered in the file IDs 4221 and 4224.

  Corresponding position offsets 4222 and 4225 register offset information in which the character string of the keyword appears in the file. In these fields 4222 and 4225, when a plurality of locations appear in one file, a plurality of offset information is registered.

  The weighting coefficients 4223 and 4226 register the importance of the appearance of the keyword character string in the file. The importance value can be appropriately set by the search server. The greater the importance value, the more important it is. The importance value can also be used to narrow down search results and to arrange search results.

  In the position information 4220, a plurality of keywords 4210 can be registered. Accordingly, it is possible to cope with a case where a plurality of files corresponding to the keyword character string exist. Note that a null value indicating that the value of the corresponding entry is invalid can be registered in the position information 4220 column. In the figure, the null value is indicated as “−”. The null value is used, for example, for an entry in which an item becomes empty because the number of registrations is smaller than other entries.

  FIG. 10 shows a configuration example of the search server management table 4300. The search server management table 4300 manages list information of search servers that are transmission destinations of search requests when the search server performs an integrated search. Specifically, the search server management table 4300 manages a search server ID 4310, a search server name 4320, an IP address 4330, and a weighting coefficient 4340 in association with each other.

  The search server ID 4310 stores an identification number for identifying a search server that can be used for the integrated search. The search server ID 4310 may be a serial number given by the search server 1100 that performs the integrated search, or may be a serial number given in the system that provides the integrated search service.

  The search server name 4320 stores the name of the search server. Specifically, it may be the host name of the search server or a name consisting of an arbitrary character string. The IP address 4330 stores the IP address assigned to the search server. In the case of a system configuration in which an IP address is determined using DNS, the IP address 4330 column may store a host name used for inquiries to DNS.

  In the weighting coefficient 4340, a value indicating the degree of importance of the search result obtained from the search server is stored. The greater the value of the weighting coefficient, the higher the importance of the search result.

  By changing the value of the weighting coefficient 4340 for each search server, the search results obtained from a specific search server can be prioritized among the integrated search results. That is, a search result from a search server having a large weighting coefficient can be displayed at the top in the integrated search result. Search results from the search server set with a small weighting coefficient are displayed in the lower rank in the ranking of the integrated search results. If the search results obtained from all the search servers are to be treated equally, all the values of the weighting coefficient 4340 may be set to the same value.

  FIG. 11 shows a configuration example of the integrated search result temporary storage table 4400. The integrated search result temporary storage table 4400 is used to temporarily store data in the process of generating the integrated search result by merging the search results from the search servers 1100, 1200, and 1300.

  Specifically, the integrated search result temporary storage table 4400 associates the search server ID 4410, rank 4420, file ID 4430, score value 4440, file path name 4450, hash algorithm 4460, hash value 4470, and search keyword-containing character string 4480. Manage.

  The search server ID 4410 stores information for identifying the search server that acquired the search result. In the search server ID 4410, the same information as the search server ID registered in the search server ID 4310 column of the search server management table 4300 is registered.

  The rank 4420 stores the rank information of the entry sent from the search server as it is. The rank is a value obtained by sorting the search results provided by each search server in the descending order of the degree corresponding to the search keyword and the search condition, and ranking in the sort order.

  The file ID 4430 stores the file ID of the file corresponding to the entry sent from the search server as it is. Specifically, the same information as the file ID registered in the file ID 4110 column of the search index registration file management table 4100 is registered in the file ID 4430.

  The score value 4440 stores the score value information of the entry sent from the search server as it is. The score value is a numerical value of the degree corresponding to the search keyword and the search condition in the search results provided by each search server. The integrated score value is calculated by multiplying the score value by the weighting coefficient 4340 in the search server management table 4300. The search server 1100 uses the integrated score value to determine the integrated rank in the integrated search result.

  The file path name 4450 stores the file path name of the file corresponding to the entry sent from the search server as it is. Specifically, in the file path name 4450, the same information as the file path name registered in the column of the acquisition source file path name 4120 of the search index registration file management table 4100 is registered.

  In addition, in order to make the target file accessible via the network 100, the file path name 4450 column may store the identification information of the file server that stores the target file in addition to the file path name. .

  The hash algorithm 4460 stores information for identifying a hash algorithm that can be used by the search server. The hash value 4470 stores a hash value calculated by a hash algorithm.

  In the negotiation process for determining the hash algorithm (common hash algorithm) used for the integrated search, if a common hash algorithm cannot be selected, an invalid value is displayed in the fields of the hash algorithm 4460 and the hash value 4470. A null value that means that is stored.

  In the search keyword-containing character string 4480, the character string containing the search keyword sent from the search server is stored as it is. The search keyword-containing character string is a set of character strings including the search keyword extracted from each file included in the search result from each search server.

  By including the information on the character string including the search keyword in the search result, the user can use the sentence or the character string of the part including the specified search keyword as a part of the search result. Thus, the user can grasp the context before and after the search keyword without actually accessing the target file listed in the search result. Therefore, the convenience of the search service can be enhanced by the search keyword-containing character string 4480.

  When there are a plurality of locations including the search keyword in one file, a plurality of search keyword-containing character strings are also registered in the column 4480. The search server uses the information registered in the search index management table 4200 to generate a search keyword-containing character string. In the column of search keyword-containing character string 4480, a null value indicating an invalid value is stored in a place that is blank because the number of search keyword-containing character strings provided from the search server is smaller than other entries. To do.

  FIG. 12 shows an example of the configuration of the integrated search request parameter 6100 specified when a client machine makes a search request to the search server 1100. This parameter is used in S1 described with reference to FIG. Specifically, the integrated search request parameter 6100 includes request destination machine identification information 6110, request source machine identification information 6120, processing type 6130, search keyword 6140, search option 6150, and integrated search option 6160.

  The request destination machine identification information 6110 stores information for identifying the search server that is the transmission destination of the integrated search request. The request destination machine identification information 6110 stores access information such as the host name or IP address of the search server in order to access the search server via the network 100.

  The request source machine identification information 6120 stores information for identifying a client machine that requests an integrated search. The request source machine identification information 6120 stores access information such as the host name of the client machine or the IP address of the client machine in order to access the client machine via the network 100.

  The process type 6130 stores information for identifying the content of the process. When issuing an integrated search request, the process type 6130 stores information indicating the integrated search request process. The search keyword 6140 stores a search keyword used for the integrated search request.

  The search option 6150 stores information on options specified when requesting a search from each search server. As the search option 6150, for example, conditions relating to file creation date / time, file update date / time, file creator and the like can be designated.

  The integrated search option 6160 stores information related to options specified for the search server 1100 that performs the integrated search processing. As the integrated search option 6160, for example, there are the number of integrated search results provided to the client machine, conditions regarding the offset value in the first entry of the integrated search results, and the like. By setting the offset value, for example, it is possible to set whether the head entry starts from rank 1 or rank 100.

  FIG. 13 shows a configuration example of the integrated search result response parameter 6200 specified when the search server 1100 returns an integrated search result to the client machine. This parameter 6200 is used in S9 described with reference to FIG. Specifically, the integrated search result response parameter 6200 includes response destination machine identification information 6210, response source machine identification information 6220, processing type 6230, processing result identification information 6240, total number 6250, response number 6260, top rank 6270, search result 6280, information 6290 necessary for an additional response request is provided.

  The response destination machine identification information 6210 stores information for identifying the client machine that is the transmission destination of the integrated search result. For example, in order to access a client machine via the network 100, access information such as a host name or an IP address of the client machine is stored.

  The response source machine identification information 6220 stores information for identifying the search server 1100 that made the integrated search request. Similarly to the above, for example, the host name and IP address of the search server 1100 are stored.

  The process type 6230 stores information for identifying the contents of the process. When the result of the integrated search is transmitted, the process type 6230 stores information indicating the response process of the integrated search result. The processing result identification information 6240 stores information for identifying the processing result of the integrated search. Specifically, information such as whether the processing has succeeded or failed has been stored.

  The total number 6250 stores the total number of file data that matches the specified condition. The response number 6260 stores the number of responses included in the response of the integrated search result among the file data matching the specified condition. When the value of the total number of cases 6250 is less than or equal to the upper limit of the number of responses 6260, the total number 6250 and the number of responses 6260 match. However, when the total number of cases 6250 is larger than the upper limit value, the amount exceeding the upper limit value of the number of response cases 6260 is not included in the response of the integrated search result.

  The head rank 6270 stores the rank value of the head entry included in the response of the integrated search result. If the entry ranked first is the first, 1 is stored in the first rank 6270, and if the entry ranked 100 is the first, 100 is stored in the first rank 6270.

  The search result 6280 stores the integrated search result acquired by the integrated search process. Search result entries 6281 and 6282 are stored in the search result 6280 as many as the number specified in the response number 6280. The search result entries 6281 and 6282 store the same information as the information stored in the respective columns 4410 to 4480 of the search result temporary storage table 4400 for integration.

  The information 6290 necessary for the additional response request is used when the value of the response number 6260 is smaller than the value of the total number 6250. In the column of information 6290 necessary for the additional response request, link information for acquiring information related to other search results not included in the response of the integrated search result is stored.

  FIG. 14 shows a hash algorithm query request for designating a query when a hash algorithm is negotiated from the search server 1100 that receives the federated search request to the search servers 1100, 1200, and 1300 that can use the federated search. The structural example of the parameter 6300 is shown.

  This parameter 6300 is used in S2 described with reference to FIG. Specifically, the hash algorithm query request parameter 6300 includes query destination machine identification information 6310, query source machine identification information 6320, processing type 6330, available hash algorithm candidate identification information 6340, and query option 6350.

  The inquiry destination machine identification information 6310 stores information for identifying a search server that is a transmission destination of a search request. In other words, the inquiry-destination machine identification information 6310 stores information for identifying each search server that needs to negotiate the hash algorithm to be used before starting the integrated search. For example, in order to access the search server via the network 100, access information such as the host name and IP address of the search server is stored.

  The inquiry source machine identification information 6320 stores information for identifying the search server 1100 that performs the integrated search process. In order to access a machine via the network 100, access information such as the host name or IP address of the search server 1100 is stored in the inquiry source machine identification information 6320.

  The process type 6330 stores information for identifying the content of the process. When making a hash algorithm inquiry, the process type 6330 stores information indicating a hash algorithm inquiry request process.

  The hash algorithm candidate identification information 6340 that can be used stores a list of identification information of hash algorithms that can be used in the search server 1100 that is the query source. In each search server, when a common hash algorithm can be used among a plurality of hash algorithms stored in the hash algorithm candidate identification information 6340, duplication included in the integrated search result can be detected using the hash algorithm.

  The query option 6350 stores option information that can be specified in the hash algorithm query request processing. Specifically, as a condition for selecting an available hash algorithm candidate, if the size of the hash value must be greater than or equal to a predetermined size, a lower limit value of the hash value size can be specified as an option.

  FIG. 15 shows a configuration example of the hash algorithm query response parameter 6400 used when each search server 1100, 1200, 1300 responds to the search server 1100 that is the hash algorithm query request source.

  This parameter 6400 is used in S3 described with reference to FIG. Hash algorithm query response parameter 6400 includes response destination machine identification information 6410, response source machine identification information 6420, processing type 6430, processing result identification information 6440, mutually usable hash algorithm identification information 6450, and usable hash algorithm candidate identification information. Includes 6460.

  The response destination machine identification information 6410 stores information for identifying the search server 1100 to which an inquiry regarding the hash algorithm should be answered. Similarly to the above, access information such as the host name or IP address of the search server 1100 is stored.

  The response source machine identification information 6420 stores information for identifying each search server that has received an inquiry about the hash algorithm. Similarly to the above, access information such as the host name or IP address of each search server is stored.

  The process type 6430 stores information for identifying the contents of the process. The processing type 6430 stores information indicating that it is a response to the hash algorithm query. The processing result identification information 6440 stores information indicating the processing result for the hash algorithm query. Specifically, the processing result identification information 6440 stores information indicating whether the inquiry processing has succeeded or failed.

  The mutually usable hash algorithm identification information 6450 stores information for identifying a hash algorithm that can be used also by the search server that has received the query, among a plurality of hash algorithms included in the usable hash algorithm candidate identification information 6340. Is done.

  Since the hash algorithm stored in the mutually usable hash algorithm identification information 6450 can be used by both the query source search server and the query destination search server, it is one of the candidates that can be used to detect duplication of integrated results. . A hash algorithm common to all search servers among the mutually usable hash algorithms returned from each search server can be selected as a hash algorithm for eliminating duplication from the integrated search result.

  The usable hash algorithm candidate identification information 6460 stores information for identifying the hash algorithm when there is another hash algorithm that can be used in the search server that has received the hash algorithm query. If the search server participating in the federated search can use a hash algorithm other than the hash algorithm that can be used by the search server 1100 that conducts the federated search (the hash algorithm registered in the column 6340 in FIG. 14), the hash algorithm Is registered in column 6460.

  The hash algorithm candidate identification information 6460 that can be used does not store the hash algorithm identification information stored in the mutually usable hash algorithm identification information 6450.

  FIG. 16 shows a configuration example of a search request parameter 6500 that is specified when the search server 1100 that has received the integrated search request issues a search request to the search servers 1100, 1200, and 1300. This parameter 6500 is used in S4 described with reference to FIG. Search request parameter 6500 includes request destination machine identification information 6510, request source machine identification information 6520, process type 6530, search keyword 6540, and search option 6550.

  The request destination machine identification information 6510 stores information (host name or IP address) for identifying the search server that is the transmission destination of the search request. The request source machine identification information 6520 stores information (host name or IP address) for identifying the search server 1100 that issues the search request.

  The process type 6530 stores information for identifying the content of the process. Here, information indicating search request processing is stored in the processing type 6530. Search keyword 6540 stores a search keyword used for the search. The search option 6550 stores option information designated for the search. For example, conditions such as file creation date / time, file update date / time, and file creator can be designated as option information.

Further, the search option 6550 includes usage hash algorithm identification information 6551. The hash algorithm identification information 6551 stores identification information of the determined hash algorithm (common hash algorithm) when a hash algorithm that is unified among the related search servers is determined by hash algorithm query processing.
Using the hash algorithm specified by the used hash algorithm identification information 6551, each search server creates a hash value and responds. Further, the search server 1100 that has received the integrated search request detects and eliminates duplicate entries from the integrated search result based on the hash value created by the common hash algorithm.

  FIG. 17 shows a configuration example of a search result response parameter 6600 that is specified when the search servers 1100, 1200, and 1300 return search results to the search server 1100 that performs integrated search. This parameter 6600 is used in S7 described with reference to FIG. Search result response parameter 6600 includes response destination machine identification information 6610, response source machine identification information 6620, process type 6630, process result identification information 6640, total number 6650, response number 6660, top rank 6670, search result 6680, additional response request Contains information needed for 6690.

  The response destination machine identification information 6610 stores information (host name or IP address) for identifying the search server that is the transmission destination of the search result. The response source machine identification information 6620 stores information (host name or IP address) for identifying the search server that has received the search request.

  The process type 6630 stores information for identifying the content of the process. Here, information indicating the search result response process is stored in the process type 6630.

  The processing result identification information 6640 stores information for identifying a search processing result. Specifically, the processing result identification information 6640 stores information indicating whether the search has succeeded or failed.

  The total number 6650 stores the total number of files and data that meet the specified condition. The number of responses 6660 stores the number of cases included in the search result response among files and data that match the specified conditions. Similarly, when the total number 6650 is equal to or less than the upper limit of the response number 6660, the total number 6650 matches the response number 6660. When the total number 6650 is larger than the upper limit value of the response number 6660, the amount exceeding the upper limit value of the response number 6660 is not included in the search result response.

  The head rank 6670 stores the rank value of the head entry for the entry included in the search result response. Similarly to the above, if the entry ranked first is the first, 1 is stored in the first rank 6670. When the entry of rank 100 is the first, 100 is stored in the first rank 6670.

  The search result 6680 stores the search result acquired by the search process. Search result entries 6680 and 6684 are stored in the search result 6680 as many as the number of responses 6680. In the search result entries 6681 and 6684, the same information as the information stored in the respective columns 4410-4480 of the search result temporary storage table 4400 for integration is stored.

  Furthermore, the search result entries 6661 and 6684 store use hash algorithm identification information 6682 and 6585 and hash values 6683 and 6686, respectively. In the used hash algorithm identification information 6682 and 6585, the information designated by the used hash algorithm identification information 6551 of the search request parameter 6500 is stored as it is.

  In the hash values 6683 and 6686, hash values created using the hash algorithm (common hash algorithm) specified by the used hash algorithm identification information 6682 and 6865 are stored. Upon receiving the integrated search request, the search server 1100 uses the hash value to detect and eliminate duplicate entries from the integrated search result.

  Information 6690 necessary for the additional response request is used when the value of the response number 6660 is smaller than the value of the total number 6650. In this case, link information for acquiring information related to the search result of the file or data not included in the search result response is stored in the column of information 6690 necessary for the additional response request.

  The search system configuration, management information configuration, and processing parameter configuration according to this embodiment have been described in detail above. Hereinafter, the processing operation according to this embodiment will be described. In the following flowchart, loops are omitted for the sake of understanding. Therefore, each flowchart shown in the figure shows an outline of each process, and is different from an actual computer program. A so-called person skilled in the art can delete or change a step from the illustrated flowchart or add a new step to the flowchart. Such a modified flowchart is also included in the scope of the present invention.

  The flowchart in FIG. 18 shows the integrated search request process executed on any client machine. First, the client machine requests an integrated search process by specifying a search keyword to the search server 1100 as an “integrated search server” that provides the integrated search service (S101). When requesting an integrated search, an integrated search request parameter 6100 is designated. After receiving the result of the integrated search from the search server 1100 that performs the integrated search process, the client machine provides the integrated search result to the user (S102), and ends this process. Note that when a response of the integrated search result is acquired from the search server 1100, the integrated search result response parameter 6200 is used.

  19 and 20 show a flowchart of the integrated search process executed by the search server 1100. FIG. First, the search server 1100 determines whether or not an integrated search request is specified based on the processing type 6130 of the integrated search request parameter 6100 received from the client machine (S201). If the integrated search request is not specified (S201: NO), the process ends with an error (S202).

  When the integrated search request is specified (S201: YES), the search server 1100 specifies a hash algorithm that can be used by the search server 1100 (S203). Specifically, by examining the hash algorithms 4151 and 4153 in the search index registration file management table 4100 managed by the search server 1100, a hash algorithm that can be used by the search server 1100 can be specified.

  The search server 1100 inquires each search server registered in the search server management table 4300 about a hash algorithm that can be used by each search server (S204). The search server 1100 designates the hash algorithm query request parameter 6300 at the time of the query.

  The search server 1100 acquires information included in the hash algorithm query response parameter 6400 from each search server. The search server 1100 determines whether or not a unified hash algorithm can be used (S205). Based on the response from each search server, the search server 1100 determines whether there is a hash algorithm that can be used in a unified manner among all search servers participating in the integrated search.

  If a unified hash algorithm can be used (S205: YES), the search server 1100 requests each search server participating in the integrated search by designating the hash algorithm to be used (S206). When requesting a search, the search server 1100 specifies a search request parameter 6500. The search server 1100 acquires information included in the search result response parameter 6600 from each search server.

  When the hash algorithm unified between the search servers participating in the federated search cannot be used (S205: NO), the search server 1100 requests the search without specifying the hash algorithm (S207). . The search server 1100 specifies a search request parameter 6500 when requesting a search. The search server 1100 acquires information included in the search result response parameter 6600 from each search server.

  Turning to FIG. After acquiring the search results, the search server 1100 stores the acquired search results in the integrated search result temporary storage table 4400 (S208). The search server 1100 determines whether duplicate entries can be excluded from the integrated search result using the hash value (S209).

  If the duplicate entry cannot be excluded from the integrated search result (S209: NO), S210 is skipped and the process proceeds to S211. When the duplicate entry can be eliminated from the integrated search result (S209: YES), the search server 1100 uses the hash value calculated by the unified hash algorithm to detect and eliminate the duplicate entry from the integrated search result ( S210).

  The search server 1100 uses the information registered in the integrated search result temporary storage table 4400, arranges the search results according to the score values, etc., and extracts entries to be provided to the integrated search request source as the integrated search results (S211).

  Specifically, the search server 1100 integrates using the score value 4440 registered in the search result temporary storage table 4400 for integration and the value of the weighting coefficient 4340 registered in the search server management table 4300. A score value is calculated. The search server 1100 uses the integrated score value to align the integrated search result entries.

  Finally, the search server 1100 returns the integrated search result to the client machine that is the request source of the integrated search (S212). The search server 1100 responds to the client machine with the integrated search result by specifying the integrated search result response parameter 6200.

  FIG. 21 is a flowchart of a response process for a hash algorithm query executed by each search server participating in the integrated search. This process is performed by each of the search servers 1100, 1200, and 1300 as “predetermined search servers”. For convenience, the search server 1200 will be described below as an example.

  First, the search server 1200 determines whether “hash algorithm query request” is specified based on the processing type 6330 specified by the hash algorithm query request parameter 6300 (S301). If the hash algorithm inquiry request is not specified (S301: NO), the process ends in error (S302).

  When a hash algorithm inquiry request is specified (S301: YES), the search server 1200 identifies a hash algorithm that can be used by the search server 1200 (S303). In this case, the “own device” in S303 is the search server 1200. The search server 1200 identifies the hash algorithms that can be used by the search server 1200 by examining the hash algorithms 4151 and 4153 of the search index registration file management table 4100 managed by the search server 1200.

  The search server 1200 determines whether there is a hash algorithm that can be used by the search server 1200 among hash algorithms that can be used by the search server 1100 that is the inquiry source (S304). The search server 1200 compares the hash algorithm specified in the hash algorithm candidate identification information 6340 that can be used in the hash algorithm query request parameter 6300 with the hash algorithm (S303) that can be used in the search server 1200. Check whether there is a common hash algorithm.

  If there is a common hash algorithm (S304: YES), the search server 1200 registers the identification information of the hash algorithm in the mutually usable hash algorithm identification information 6450 in the hash algorithm query response parameter 6400 ( S305).

  If there is no common hash algorithm between the hash algorithm that can be used in the query server 1100 and the query algorithm that can be used in the query server 1200 (S304: NO), S305 is skipped and S306 is skipped. Move on.

  The search server 1200 determines whether there is another hash algorithm that can be used by the search server 1200 other than the mutually usable hash algorithm discovered in S304 (S306). The search server 1200 determines whether or not there is another hash algorithm that has not been registered in the process S305 among the hash algorithms identified as the hash algorithm that can be used by the search server 1200 in the process S303. Investigate.

  If another hash algorithm exists (S306: YES), the search server 1200 registers the identification information of the hash algorithm in the available hash algorithm candidate identification information 6460 in the hash algorithm query response parameter 6400 ( S307). If there is no other hash algorithm (S306: YES), S307 is skipped and the process proceeds to S308.

  The search server 1200 returns the query result of the hash algorithm to the search server 1100 that is the query source (S308). The search server 1200 specifies the hash algorithm query response parameter 6400 and responds with the query result.

  FIG. 22 shows a flowchart of search response processing executed in each search server. This process is performed by each of the search servers 1100, 1200, and 1300, similarly to the process described in FIG. Here, for convenience, the search server 1200 will be described as an example.

  First, the search server 1200 checks the processing type 6530 specified in the search request parameter 6500 and determines whether “search request” is specified (S401). If not specified (S401: NO), this process ends in error (S402). If it is specified (S401: YES), the search server 1200 executes a search process with the specified search keyword and acquires the search result (S403). The search server 1200 uses the search keyword 6540 and the search option 6550 in the search request parameter 6500 to perform a search process.

  The search server 1200 checks whether the usable hash algorithm identification information 6551 is specified in the search option 6550 of the search request parameter 6500 (S404). If not specified (S404: NO), S405 is skipped and the process proceeds to S406.

  If specified (S404: YES), the search server 1200 adds the hash value of the file included in each entry and the hash algorithm identification information used to generate the hash value to each entry of the acquired search results. Register (S405). The search server 1200 acquires a hash value and hash algorithm identification information based on information stored in the hash algorithm 4150 of the file registered in the search index registration file management table 4100.

  The search server 1200 returns the search result to the search server 1100 as the request source (S406). The search server 1200 specifies the search result response parameter 6600 and responds with the search result.

  FIG. 23 shows a flowchart of search index update processing. This processing is performed by each search server 1100, 1200, 1300, respectively. For convenience, the search server 1200 will be described below as an example.

  First, the search server 1200 specifies a hash algorithm that can be used by the search server 1200 (S501). The “own device” in S501 is the search server 1200 here. The search server 1200 identifies hash algorithms that can be used by the search server 1200 by examining the hash algorithms 4151 and 4153 in the search index registration file management table 4100 managed by the search server 1200.

  The search server 1200 identifies the file server that is the update target of the search index and the root directory that is the update target (S502). Next, the search server 1200 crawls all search index update target files and determines whether or not the indexing is completed (S503). When crawling or the like has been completed for all files (S503: YES), this process ends.

  When the crawling and indexing processes for all files have not been completed (S503: NO), the search server 1200 accesses the file server in which the crawling target file is stored, and is stored in the search index update target range. One arbitrary file is acquired (S504).

  The search server 1200 determines whether information related to the file acquired in S504 needs to be newly registered in the search index or whether information related to the file acquired in S504 needs to be updated on the search index ( S505).

  Specifically, the search server 1200 determines whether the acquired file has been updated since the previous search index update process or whether the acquired file has been newly stored after the previous search index update process. Examine the viewpoint of whether or not there is. If new registration or update is not required (S505: NO), the process returns to S503. When new registration or update is necessary (S505: YES), the process proceeds to S506 shown in FIG. The search server 1200 determines whether the information of the file (target file) acquired in S504 is newly registered in the search index, or information already registered in the search index is updated (S506).

  If it is determined that new registration is to be performed, the search server 1200 newly creates an entry for the target file in the search index registration file management table 4100 and registers information about the target file (S507).

  When it is determined that the update is to be performed, the search server 1200 identifies the entry of the target file stored in the search index registration file management table 4100, and updates necessary information (S508). The search server 1200 analyzes the target file and registers the search index information in the search index management table 4200 (S509). The search server 1200 confirms whether there is an available hash algorithm (S510). The search server 1200 determines whether one or more hash algorithms that can be used by the search server 1200 exist based on the identification result of S501. If there is no available hash algorithm (S510: NO), the process returns to S503.

  If there is an available hash algorithm (S510: YES), the search server 1200 generates each hash value from the data of the target file using all the available hash algorithms, and generates each hash. The value is registered in the search index registration file management table 4100 (S511).

  When there are a plurality of hash algorithms that can be used, the search server 1200 generates hash values corresponding to all the hash algorithms and registers them in the search index registration file management table 4100.

  In the present embodiment configured as described above, when performing an integrated search in a system in which a plurality of search servers having different search algorithms or / and search index update timings, etc. are sparsely combined, Determine the hash algorithm to be used.

  Therefore, in this embodiment, duplicate entries can be detected and eliminated from the integrated search results obtained by integrating the search results from the search servers related to the same search condition. Thereby, the user can obtain an integrated search result across a plurality of search servers without waste. The user can find the target file relatively easily by using the integration result from which duplicate entries are removed, and the usability of the user is improved.

  In this embodiment, among the plurality of search servers 1100, 1200, and 1300 that are loosely coupled, the search server 1100 that has received the integrated search request negotiates with each of the search servers 1100, 1200, and 1300, whereby each search server 1100, The hash algorithm used in common between 1200 and 1300 is decided, and the actual search is performed by each search server 1100, 1200 and 1300, respectively. Furthermore, in the present embodiment, each search server 1100, 1200, 1300 creates a hash value using the determined hash algorithm, and the search server 1100 that receives the integrated search request uses the hash value to perform the integrated search result. Detect duplicate entries from and remove them. In this embodiment, the creation of a hash value is distinguished from the detection and elimination of duplication using the hash value. Thereby, in a present Example, a role can be divided among several loosely coupled search servers.

  A second embodiment will be described with reference to FIGS. Each of the following embodiments including this embodiment corresponds to a modification of the first embodiment. Accordingly, the following embodiments will be described with a focus on differences from the first embodiment.

  In the first embodiment described above, every time an integrated search process is performed between the search server 1100 that has received the integrated search request and the other search servers 1100, 1200, and 1300, duplicate entries are extracted from the integrated search results. Negotiate a hash algorithm to use to eliminate.

  However, the hash algorithm that is commonly used among the search servers 1100, 1200, and 1300 is not frequently changed. In the normal case, once determined, the same hash algorithm is considered to be used for a relatively long time.

  Therefore, in this embodiment, the hash algorithm information acquired first is stored as a cache in the search server 1100 that accepts the integrated search request. Thereafter, when an integrated search request is issued, a common hash algorithm is determined using cached hash algorithm information, and an integrated search process is executed. Therefore, in this embodiment, it is not necessary to negotiate the hash algorithm between each search server every time an integrated search request is received, and the overhead at the start of the integrated search can be reduced.

  In order to cache in the search server 1100 one or more hash algorithms respectively acquired from each search server 1100, 1200, 1300, the configuration of the search server management table 4300 that the search server 1100 has, It is necessary to change a part of the integrated search process executed by the search server 1100.

  FIG. 25 is a configuration example of the search server management table 4300 used in this embodiment. In the search server management table 4300, in addition to the columns 4310, 4320, 4330, and 4340 described in FIG. 10, a column for managing identification information 4350 of the available hash algorithm is added.

  The use hash algorithm identification information 4350 stores information for identifying hash algorithms that can be used by the search servers 1100, 1200, and 1300 participating in the integrated search. A plurality of hash algorithm identification information can be stored for one search server. For example, in FIG. 25, SHA-1 and SHA-2 are stored as the use hash algorithm identification information 4350 in the entry with the search server ID 4310 as the first. In the integrated search process, the identification information of the hash algorithm that can be used is registered in the table 4300 based on the information of the available hash algorithm acquired from each search server.

  FIG. 26 shows the change contents of the integrated search process executed by the search server 1100. This process differs from the integrated search process shown in FIG. 19 in the following points. The first difference is that after S203, the search server 1100 determines whether there is available hash algorithm information in the search server management table 4300 (S213). The search server 1100 confirms whether or not the hash algorithm identification information is registered in the entry of the used hash algorithm identification information 4350 in the search server management table 4300.

  When the usable hash algorithm information is stored in the search server management table 4300 (S213: YES), the search server 1100 skips the hash algorithm negotiation process S204, and proceeds to S205. If the usable hash algorithm information is not stored in the search server management table 4300 (S213: NO), the search server 1100 proceeds to S204 to perform a hash algorithm negotiation process as in the first embodiment.

  The second difference is that after S204, the search server 1100 registers the hash algorithm identification information acquired in S204 in the search server management table 4300 (S214). Specifically, the search server 1100 stores the hash algorithm identification information acquired from each of the other search servers 1100, 1200, and 1300 in the column of the used hash algorithm identification information 4350. When a plurality of hash algorithm identification information is acquired for one search server, all the hash algorithm identification information is registered in the search server management table 4300. After S214, the process proceeds to S205.

  Configuring this embodiment like this also achieves the same effects as the first embodiment. Further, in this embodiment, the hash algorithm identification information acquired at the time of the first integrated search is held, a common hash algorithm is determined, and the integrated search is performed. Therefore, it is not necessary to acquire information about the hash algorithm from each search server 1100, 1200, 1300 each time an integrated search request is received, and the overhead of the integrated search can be shortened.

  A third embodiment will be described with reference to FIGS. In the first embodiment described above, the hash algorithm is negotiated between the search server 1100 that receives the integrated search request and the other search servers 1100, 1200, and 1300 each time the integrated search process is performed. However, as described in the second embodiment, the hash algorithm is not frequently changed. Therefore, in this embodiment, as described below, when a system that provides an integrated search service is constructed, the search server 1100 acquires hash algorithms that can be used by the search servers 1100, 1200, and 1300, respectively. Register in advance in server 1100.

  The search server management table 4300 needs to be changed in the same way as the contents described in FIG. 25 of the second embodiment. The content of the change is the same as in FIG. For the integrated search control program 1125 in the search server 1100, a process for registering a hash algorithm in advance is newly added.

  FIG. 27 shows the configuration of a computer program that the search server 1100 has. 27, in addition to the configuration shown in FIG. 3, a hash algorithm pre-negotiation subprogram 1178 is newly added to the integrated search control program 1125.

  The hash algorithm pre-negotiation subprogram 1178 investigates in advance the hash algorithm used in each of the search servers 1100, 1200, and 1300 at the time of system construction for providing the integrated search service, and the search result is obtained from the search server 1100. It is a process for making it store in.

  FIG. 28 shows a flowchart of the hash algorithm pre-negotiation process executed by the search server 1100. This process is performed when setting the search servers 1100, 1200, and 1300 in the case of constructing a system that provides an integrated search service.

  First, the search server 1100 identifies a hash algorithm that can be used by the search server 1100 (S601). It can be specified by examining the hash algorithms 4151 and 4153 in the search index registration file management table 4100 managed by the search server 1100.

  The search server 1100 inquires of all the search servers 1100, 1200, 1300 registered in the search server management table 4300 about the hash algorithm (S602). The hash algorithm query request parameter 6300 is used for this query.

  The search server 1100 acquires information included in the hash algorithm query response parameter 6400 from each of the search servers 1100, 1200, and 1300 that are the query destinations. The search server 1100 registers each hash algorithm identification information in the search server management table 4300 (S603).

  Since the present embodiment is configured as described above, the same effects as the first embodiment can be obtained. Furthermore, in this embodiment, information related to the hash algorithm is collected and stored from each search server when a system that provides an integrated search service is constructed. Therefore, it is not necessary to negotiate the hash algorithm every time an integrated search request is received, and the overhead of the integrated search can be shortened.

  A fourth embodiment will be described with reference to FIG. The first embodiment described above creates a hash value for search target file data at the time of search index update processing in each search server 1100, 1200, 1300. On the other hand, in the present embodiment, as described below, a hash value is created during the search process. Therefore, in this embodiment, the overhead of the search index update process can be reduced, and an area for storing the hash value can be made unnecessary.

  In this embodiment, when the search server receives a search request, the search server searches for file data that matches the search condition and creates a hash value for the file data on a so-called on-demand basis.

  FIG. 29 shows a flowchart of search response processing in the search server. This process includes S401 to S406 shown in FIG. 22, and S407 to S409 are newly added between S404 and S405. Hereinafter, as in FIG. 22, the subject is the search server 1200.

  If the search request specifies a hash algorithm (S404: YES), the search server 1200 determines whether a hash value has been created using the specified hash algorithm (S407). The search server 1200 checks whether or not the hash value of the target file is registered in the field of the hash algorithm 4150 of the target file in the search index registration file management table 4100.

  When the hash value has been created (S407: YES), the process proceeds to S405. On the other hand, when the hash value has not been created (S407: NO), the search server 1200 acquires the target file data (S408). The search server 1200 may use file data acquired by crawling for updating the search index, or may acquire target file data from the file server anew.

  After acquiring the target file data, the search server creates a hash value from the target file data using a designated hash algorithm (S409). After creating the hash value, the search server 1200 moves to S405.

  Configuring this embodiment like this also achieves the same effects as the first embodiment. Furthermore, in this embodiment, when a search request is received, a hash value of a file that matches the search condition is generated on the spot. Therefore, in this embodiment, it is not necessary to create a hash value or store the hash value at the time of search index update processing.

  A fifth embodiment will be described with reference to FIG. In the fourth embodiment described above, hash values are created for the search target files during the search processing in the search servers 1100, 1200, and 1300. However, so-called on-demand hash value creation may be difficult depending on the processing load or machine performance of each search server.

  Therefore, in this embodiment, the search server 1100 that has received the integrated search request creates hash values for the files searched by the search servers 1100, 1200, and 1300, respectively. Thereby, in this embodiment, it is possible to reduce the overhead at the time of search processing in each search server 1100, 1200, 1300, or to reduce the area for storing hash values.

  FIG. 30 shows a part of a flowchart of integrated search processing in the search server 1100. The flowchart corresponds to the flowchart shown in FIG. I will focus on the differences.

  The search server 1100 acquires the file data stored in the entry of the integrated search result temporary storage table 4400 (S213). The search server 1100 may directly acquire file data from the file server based on the file path name 4450 of the integrated search result temporary storage table 4400. Alternatively, when cache data of the target file is stored in the search server 1100, the cache data may be used.

  The search server 1100 creates a hash value of each file data acquired in S213 using a hash algorithm that can be commonly used by each search server, and registers it in the integrated search result temporary storage table 4400 (S214). The search server 1100 stores the identification information of the common hash algorithm and the created hash value in the hash algorithm 4460 column and the hash value 4470 column in the integrated search result temporary storage table 4400.

  Configuring this embodiment like this also achieves the same effects as the first embodiment. Further, in this embodiment, when receiving the integrated search request, the search server 1100 generates a hash value of the file that matches the search condition on the spot. Therefore, in this embodiment, when it is difficult to create a hash value in each search server because the processing load becomes too high, it is possible to create a hash value collectively at the search server that has received the integrated search request.

  Furthermore, in this embodiment, the share of hash value creation can be changed according to the load status of each search server. For example, a search server with a low load creates a hash value within the search server, and a search server with a high load creates a hash value with the search server (or other search server) that received the integrated search request. You can also

  A sixth embodiment will be described with reference to FIGS. In the fifth embodiment described above, during the integrated search process in the search server 1100, the search server 1100 creates a hash value of the target file data.

  However, the search server 1100 that receives the integrated search request may not be able to access the file server storing the target file data. In this case, the search server 1100 that has received the integrated search request cannot acquire file data that matches the search conditions.

  Therefore, instead of using file data, a character string containing a search keyword provided by each search server as a part of the search result is used. In this embodiment, a hash value is created based on the search keyword-containing character string, and is used to detect and eliminate duplicate entries from the integrated search result. As a result, when the search server 1100 that receives the integrated search request cannot access the file server, or when it is difficult to create a hash value on each search server 1100, 1200, 1300 due to high processing load, the integrated search Duplicate entries can be found and eliminated from the results.

  FIG. 31 shows an integrated search result temporary storage table 4400. In the integrated search result temporary storage table 4400 of FIG. 31, a partial character string 4481 and a partial hash value 4482 are newly added to the search keyword-containing character string 4480, compared to the integrated search result temporary storage table 4400 shown in FIG. Is different.

  The partial character string 4481 is information originally stored in the search keyword-containing character string 4480 in the integrated search result temporary storage table 4400 shown in FIG. In the partial hash value 4482, a hash value created from the partial character string 4482 is stored.

  The partial hash value 4482 may be created using a hash algorithm registered in the hash algorithm 4440 of the integrated search result temporary storage table 4400. Alternatively, an arbitrary hash algorithm that can be used by the search server 1100 may be selected and created using the hash algorithm.

  A plurality of sets of partial character strings 4482 and partial hash values 4482 stored in the search keyword-containing character string 4480 can be stored for each entry. If a blank is generated in the partial character string 4482 or the partial hash value 4482 in the search keyword-containing character string 4480 field, a null value may be stored.

  FIG. 32 shows a part of a flowchart of the integrated search process executed by the search server 1100. In this process, a hash value for the search keyword-containing character string included in the search result is created on demand.

  In FIG. 32, after S208, the search server 1100 creates a hash value of the partial character string stored in the entry of the integrated search result temporary storage table 4400 using a hash algorithm that can be used by the search server 1100. It is registered in the integrated search result temporary storage table 4400 as a partial hash value (S220).

  Here, in the entry of the integrated search result temporary storage table 4400, when a plurality of partial character strings are stored in the partial character string 4481, the search server 1100 creates partial hash values for all the partial character strings, respectively. And stored in the partial hash value 4482 column.

  The search server 1100 detects a duplicate entry from the integrated search result using the partial hash value, and eliminates it (S221).

  The search server 1100 can determine that an entry in which all partial hash values match is a duplicate entry. Alternatively, an entry having a partial hash value that matches a certain ratio or more may be determined to be a quasi-duplicate entry. For example, two entries having m (0 <m <n) or more partial hash values that match among the n partial hash values may be determined to be duplicate entries.

  Configuring this embodiment like this also achieves the same effects as the first embodiment. Further, in this embodiment, the hash value is not obtained from the file data, but the hash value is obtained from the character string including the search keyword (that is, a part of the file data) and used for detecting the duplicate entry. Therefore, even if the search server 1100 cannot access the file server, or even if a common hash algorithm cannot be determined between the search servers 1100, 1200, and 1300, duplicate entries are found and eliminated from the integrated search results. be able to.

  A seventh embodiment will be described with reference to FIG. In this embodiment, when a duplicate entry is detected from the integrated search result, the search server that manages the search index corresponding to the duplicate entry is notified that there is a duplicate entry.

  FIG. 33 shows processing when a duplicate entry is found. The representative search server 1100 is a search server for receiving an integrated search request and performing an integrated search. The search servers (1100, 1200, 1300) are search servers that participate in the integrated search, and search according to a search request from the representative search server 1100.

  Upon receiving the integrated search request from the client machine, the representative search server 1100 issues a search request to each search server (S701). Each search server searches according to the search request, and returns the search result to the representative search server 1100 (S702). The representative search server 1100 detects a duplicate entry from the integrated search result using the hash value (S703).

  The representative search server 1100 transmits the integrated search result from which the duplicate entry is removed to the client machine (S704). Further, the representative search server 1100 notifies the search server in which the duplicate entry has been found that the duplicate entry has been found (S705). Upon receiving the notification, the search server confirms that it is a duplicate entry (S706). The search server can instruct the file server to delete the file corresponding to the duplicate entry.

  Configuring this embodiment like this also achieves the same effects as the first embodiment. Furthermore, in this embodiment, when a duplicate entry is detected from the integrated search result, a notification to that effect is sent to the search server, so that a duplicate file can be deleted.

  In addition, this invention is not limited to embodiment mentioned above. A person skilled in the art can make various additions and changes within the scope of the present invention. For example, when a hash algorithm common to a plurality of search servers is not found, a configuration may be adopted in which a hash algorithm that can be a common hash algorithm is transmitted and installed in a search server that does not have a hash algorithm that can be a common hash algorithm. .

Claims (15)

A method of searching using a computer system including a plurality of search servers,
The computer system is configured by loosely coupling the plurality of search servers that operate independently of each other,
When the integrated search server included in the plurality of search servers receives an integrated search request for causing the plurality of predetermined search servers included in the plurality of search servers to search, respectively, the predetermined search servers are commonly used. Determining available duplicate detection information for detecting duplicates, issuing a search request corresponding to the integrated search request to each predetermined search server;
Each of the predetermined search servers searches a data group in charge based on the search request, and a duplicate detection for detecting duplicates created using the determined duplicate detection information in the search result. Including the value to the federated search server,
The integrated search server detects duplicate data based on the duplicate detection values from the search results received from the predetermined search servers, and detects the duplicate data detected from the search results. Removing and creating an integrated search result, and providing the integrated search result to the issuer of the integrated search request;
retrieval method.
Each of the predetermined search servers is
For the data group in charge, each of the duplicate detection values for each of a plurality of duplicate detection information is stored in advance,
The duplicate detection value corresponding to the duplicate detection information determined by the integrated search server among the stored duplicate detection values is included in the search result and transmitted to the integrated search server. Search method.
Each of the predetermined search servers creates and stores the duplicate detection value for each of the plurality of duplicate detection information when updating a search index used for searching the data group in charge. Item 3. The search method according to Item 2.
The integrated search server acquires and holds information on duplicate detection information that can be used by each predetermined search server from each predetermined search server, and holds the information when the integrated search request is received. The search method according to claim 1, wherein the duplicate detection information that can be commonly used by the predetermined search servers is determined based on the information related to each duplicate detection information.
When the computer system is constructed, the integrated search server acquires information about duplicate detection information that can be used by each predetermined search server from the predetermined search server, and holds the information. 2. The search according to claim 1, wherein, when an integrated search request is received, the duplicate detection information that can be commonly used by each of the predetermined search servers is determined based on the held information relating to each duplicate detection information. Method.
When each of the predetermined search servers receives the search request from the integrated search server, the predetermined search server creates the duplicate detection value based on the determined duplicate detection information, and includes the duplicate detection value in the search result. The search method according to claim 1, wherein the search method is transmitted to the integrated search server.
The search method according to claim 1, wherein the duplicate detection information is a hash algorithm, and the duplicate detection value is a hash value.
An integrated search server for searching using a computer system configured by loosely coupling a plurality of search servers that operate independently of each other,
When receiving an integrated search request for causing each of a plurality of predetermined search servers included in the plurality of search servers to search,
The duplication detection information for detecting duplication that can be used in common by each of the predetermined search servers is determined,
Specify the determined duplication detection information to each predetermined search server, issue a search request corresponding to the integrated search request,
Each of the predetermined search servers is a search result obtained by searching a data group in charge, and the search result including a duplicate detection value for detecting a duplicate created using the determined duplicate detection information. , Received from each of the predetermined search servers,
From the search results received from each predetermined search server, detect duplicate data based on each duplicate detection value,
Remove the detected duplicate data to create an integrated search result,
Providing the integrated search result to an issuer of the integrated search request;
Integrated search server.
The integrated search server acquires and holds information on duplicate detection information that can be used by each predetermined search server from each predetermined search server, and holds the information when the integrated search request is received. Based on the information on each duplicate detection information that has been made, determine the duplicate detection information that can be commonly used by each of the predetermined search servers,
The integrated search server according to claim 8.
When the computer system is constructed, the integrated search server acquires information about duplicate detection information that can be used by each predetermined search server from the predetermined search server, and holds the information. 9. The integration according to claim 8, wherein, when an integrated search request is received, the duplication detection information that can be commonly used by each of the predetermined search servers is determined based on the held information about each duplication detection information. Search server.
When each of the predetermined search servers receives the search request from the integrated search server, the predetermined search server creates the duplicate detection value based on the determined duplicate detection information, and includes the duplicate detection value in the search result. The federated search server according to claim 8, wherein the federated search server is transmitted to the federated search server.
A computer program for causing a computer to function as an integrated search server for searching using a computer system configured by loosely coupling a plurality of search servers that operate independently of each other,
In the computer,
Receiving an integrated search request for causing each of a plurality of predetermined search servers included in the plurality of search servers to search;
The duplication detection information for detecting duplication that can be commonly used by each of the predetermined search servers is determined,
Each of the predetermined search servers, specifying the determined duplication detection information, causing a search request corresponding to the integrated search request to be issued,
Each of the predetermined search servers is a search result obtained by searching a data group in charge, and the search result including a duplicate detection value for detecting a duplicate created using the determined duplicate detection information. , Received from each of the predetermined search servers,
From the search results received from each predetermined search server, to detect duplicate data based on each duplicate detection value,
Remove the detected duplicate data and create an integrated search result,
Providing the integrated search result to an issuer of the integrated search request;
Computer program.
In the computer,
Information about duplicate detection information that can be used in each predetermined search server is acquired from each predetermined search server and held;
When receiving the integrated search request, based on the information on each held duplicate detection information, to determine the duplicate detection information that each predetermined search server can use in common,
The computer program according to claim 12.
In the computer,
When the computer system is constructed, information on duplicate detection information that can be used in each predetermined search server is acquired from each predetermined search server and held,
When receiving the integrated search request, based on the information on each held duplicate detection information, to determine the duplicate detection information that each predetermined search server can use in common,
The computer program according to claim 12.
The computer program according to claim 12, wherein the duplicate detection information is a hash algorithm, and the duplicate detection value is a hash value.

Images